JOSHUA R. SANES, Ph.D.
Harvard Medical School
Q: You’ve recently been appointed to head Harvard’s new multidisciplinary initiative in systems neuroscience. How does a self-described “reductionist” end up in this position?
Sanes: I’m fascinated by systems-level questions, and see them as probing a huge frontier. I would say I descended into reductionism in graduate school. My undergraduate major [at Yale] was psychology, because I was very interested in mental illness, especially schizophrenia. I hoped to find some way to study the biochemistry of mental illness, but work in that field at the time was really pretty marginal, because of the lack of appropriate technology. I was eager to become a scientist, and frankly, I wasn’t aware of a way that one could do good science on mental illness at that point. So, studying synaptic development seemed fascinating and something that was bound to be relevant in the long run. It was a way to break complicated matters, which we didn’t understand, into simpler matters, which we might be able to understand. I thought that was a good thing.
Now I have two jobs. In my own lab, we are continuing to take a fairly reductionist approach, by studying synaptic development at the neuromuscular junction. I’d say we’re upgrading in that we’re now trying to study synaptic development in the actual brain as well, because techniques have gotten good enough to do that. And we are thinking about ways to use what we have found out about development to look at the function of particular types of synapses within circuits. The systems part of my life is the second job, as director at Harvard’s new Center for Brain Sciences. What I keep telling people is that it really will be systems neuroscience, and I’d never hire myself for it. It will not be a group of people who do the sorts of things I do, but rather a group that includes real systems neuroscientists along with people in other disciplines whose work is aimed directly at systems-level issues.
Q: What is meant by a “systems neuroscientist”?
A: The broadest definition of a systems neuroscientist is someone who is interested not just in the functions of individual molecules, synapses, or cells, but also in how they work together to account for behavior. That’s not to say that we know everything about the cells and molecules in development. Those are vibrant areas, the ones in which the most clinically relevant advances are being made right now. It would be a disaster to abandon those molecular and cellular approaches; that’s where treatments for ALS, for Alzheimer’s, for multiple sclerosis are coming from. But, as a long-term investment, I think that we need to start working more on systems-level issues in parallel. They are going to be of very profound clinical relevance, but with a longer lead time. That’s all the more reason to get going now.
I’d say people like David Hubel and Torsten Wiesel, who did their Nobel Prize-winning work in the early 1960s, were absolutely paradigmatic systems neuroscientists. What happened, I think, was that progress slowed in the area over the next few decades, for technological reasons. If you want to know how a whole ensemble of neurons makes a circuit that underlies a neural computation, you could only get so far by recording from a couple neurons at a time out of the millions or billions in the circuit, and having relatively simple read-outs of what those neurons are doing. So where I think a lot of the excitement comes from, is not that the questions have changed, but that technological advances are now enabling us to ask them—and actually answer them—in many new ways.
Q: What are some of the critical technological advances that make it possible to ask systems-level questions now?
A: One of the biggest is functional MRI, which essentially lets people image mental activity in the living brain. It has the big limitation that what you’re imaging are groups of thousands or tens of thousands of cells, so you’re not getting at circuits and individual connections between the cells in the sense that I find most interesting. But it’s tremendously valuable: it lets you see which parts of the brain are involved; it lets you target further studies; it lets you put together diagrams that enable you to compare patterns of activity for different brain functions. As it is being applied now to monkeys and rats, it lets you home in on where you’d want to do the old-fashioned sort of recording. And, of course, it can be—is being— applied to people. You just can’t overestimate how big that change is.
Another set of advances is in light microscopy. These new methods make it possible to see neurons and synapses in the living brain in whole animals—mice, fish, or flies—to really look at circuits with the potential of seeing how they change when behaviors change. Light microscopy really encompasses a whole suite of individual technologies, including ways to put fluorescent proteins into individual neurons so you can see them and their connections. At the same time, advances in confocal and multi-photon microscopy enable you to focus through this big white blob that is the brain to see individual cells and connections. And then there have been huge computational advances, so you can take a gazillion images and reconstruct them to see these complicated neurons. There are also methods just coming on line that use molecular technologies to optically record the activity of neurons in living animals, by manipulating the fluorescence so that it changes color every time it fires. So you can now imagine not only seeing the circuit but seeing the activity of the circuit, and how you might affect the activity.
Then there are advances in genomics that will allow us to study animals with interesting behavior rather than just the few species for which we have traditional genetic tools. Then there are new types of multi-site recording. I could go on and on. It is really a wonderful time.
Q: What is the goal of systems neuroscience, as you see it?
A: Central to the mission at our center is to find wiring diagrams that underlie simple behaviors. We might do it in ants, honeybees, birds, or mice, whichever ones turn out to be best. Then, once we can relate the wiring diagrams to behaviors, we hope to look at how the wiring diagrams are different among individuals. We want to learn what accounts for individual differences in behavior. Which differences are genetic and which ones result from experience? This is the old nature-nurture question but posed in a new way. We also hope to find out how the wiring diagrams change to underlie changes in behavior in an individual over time, such as aging, for example.
So one fantasy would be to find a circuit that seems to underlie a behavior. It could be an ant navigating toward its nest, maternal behavior in mice, or bird-song. And you would fluorescently label, let’s say, 10, 20, 30 different types of neurons different colors so you could tell them apart and you could see their activity in the circuit. You would map the circuit, see what the activity is, and then turn off, say, neuron No. 4, and see whether it really was important or whether it was a bystander. I think in 10 years, people will be able to do that, watching it all in real time. A lot of the little pieces can be done already, but you can’t underestimate the job it will require to put those pieces together.
Q: How might systems neuroscience help solve the mysteries of mental illnesses?
A: I think that some of these illnesses, like schizophrenia, autism, learning disabilities and so on, may really be diseases of circuits. I think of it in sort of a simple-minded way, that there are wiring diagrams in the brain, similar to what engineers make when they map an electrical circuit. And in some behavioral illnesses, there is something wrong either with how the neurons are wiring up or with how information passes through these circuits. That’s not to say that molecules won’t be important; the fact that there are really good antipsychotic medicines shows that molecular approaches are critical. Yet, I think it’s likely that as people look more closely at the genes and molecules involved in these conditions, they’ll keep finding more and more. There’s no question that we need that information, but if the real underlying defect is in the processing of information—the circuitry—a really holistic understanding of it would come only from understanding what’s wrong with the circuits. If we had this sort of global understanding, we might be in a much better position to think rationally about how to help.
“Now these technologies are coming along to let systems neuroscientists fill in the gap...It is not a tiny little gap. It’s a chasm.”
Where I think systems neuroscience has a huge role to play is in that vast land between molecules at the low end, and cognitive neurosciences at the high end, in figuring out the wiring diagrams and relating them to the output, the behavior. At the high end, fMRI and so forth can tell us where the important circuits are—maybe there’s nothing wrong with the schizophrenic’s circuit in the thalamus but there is something wrong in the prefrontal cortex. You’ve got to know that before you go there to look more closely. At the very reductionist end, it may be that the circuit is not working right because gene # 4073 isn’t right. We’ve been learning a lot about the molecules, and also about cognitive processing. But there is a gap because molecules don’t cause behavior; I think electrical activity in circuits causes behavior. Now these technologies are coming along to let systems neuroscientists fill in the gap. And I think that as of now, it is mostly gap. It is not a tiny little gap. It’s a chasm.
A really critical long-term strategy, then, is to try to bridge that gap. But it really is a long-term investment. If I had a child with a mental illness, I would bank on the molecular road in the short term. But eventually, to really get it right, we will not be able to leap from a molecule past all the circuits to the cognition. We will really need to understand the steps in-between.